exec.c 52.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
/*
 *  linux/fs/exec.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

/*
 * #!-checking implemented by tytso.
 */
/*
 * Demand-loading implemented 01.12.91 - no need to read anything but
 * the header into memory. The inode of the executable is put into
 * "current->executable", and page faults do the actual loading. Clean.
 *
 * Once more I can proudly say that linux stood up to being changed: it
 * was less than 2 hours work to get demand-loading completely implemented.
 *
 * Demand loading changed July 1993 by Eric Youngdale.   Use mmap instead,
 * current->executable is only used by the procfs.  This allows a dispatch
 * table to check for several different types  of binary formats.  We keep
 * trying until we recognize the file or we run out of supported binary
 * formats. 
 */

#include <linux/slab.h>
#include <linux/file.h>
A
Al Viro 已提交
27
#include <linux/fdtable.h>
H
Hugh Dickins 已提交
28
#include <linux/mm.h>
L
Linus Torvalds 已提交
29 30
#include <linux/stat.h>
#include <linux/fcntl.h>
H
Hugh Dickins 已提交
31
#include <linux/swap.h>
32
#include <linux/string.h>
L
Linus Torvalds 已提交
33
#include <linux/init.h>
34
#include <linux/pagemap.h>
35
#include <linux/perf_event.h>
L
Linus Torvalds 已提交
36 37 38 39 40 41
#include <linux/highmem.h>
#include <linux/spinlock.h>
#include <linux/key.h>
#include <linux/personality.h>
#include <linux/binfmts.h>
#include <linux/utsname.h>
42
#include <linux/pid_namespace.h>
L
Linus Torvalds 已提交
43 44 45 46 47
#include <linux/module.h>
#include <linux/namei.h>
#include <linux/mount.h>
#include <linux/security.h>
#include <linux/syscalls.h>
48
#include <linux/tsacct_kern.h>
M
Matt Helsley 已提交
49
#include <linux/cn_proc.h>
A
Al Viro 已提交
50
#include <linux/audit.h>
R
Roland McGrath 已提交
51
#include <linux/tracehook.h>
J
Johannes Berg 已提交
52
#include <linux/kmod.h>
53
#include <linux/fsnotify.h>
54
#include <linux/fs_struct.h>
55
#include <linux/pipe_fs_i.h>
Y
Ying Han 已提交
56
#include <linux/oom.h>
57
#include <linux/compat.h>
L
Linus Torvalds 已提交
58 59 60

#include <asm/uaccess.h>
#include <asm/mmu_context.h>
61
#include <asm/tlb.h>
62 63

#include <trace/events/task.h>
64
#include "internal.h"
L
Linus Torvalds 已提交
65

66 67
#include <trace/events/sched.h>

L
Linus Torvalds 已提交
68
int core_uses_pid;
69
char core_pattern[CORENAME_MAX_SIZE] = "core";
70
unsigned int core_pipe_limit;
A
Alan Cox 已提交
71 72
int suid_dumpable = 0;

73 74 75 76 77 78
struct core_name {
	char *corename;
	int used, size;
};
static atomic_t call_count = ATOMIC_INIT(1);

L
Linus Torvalds 已提交
79 80
/* The maximal length of core_pattern is also specified in sysctl.c */

A
Alexey Dobriyan 已提交
81
static LIST_HEAD(formats);
L
Linus Torvalds 已提交
82 83
static DEFINE_RWLOCK(binfmt_lock);

I
Ivan Kokshaysky 已提交
84
int __register_binfmt(struct linux_binfmt * fmt, int insert)
L
Linus Torvalds 已提交
85 86 87 88
{
	if (!fmt)
		return -EINVAL;
	write_lock(&binfmt_lock);
I
Ivan Kokshaysky 已提交
89 90
	insert ? list_add(&fmt->lh, &formats) :
		 list_add_tail(&fmt->lh, &formats);
L
Linus Torvalds 已提交
91 92 93 94
	write_unlock(&binfmt_lock);
	return 0;	
}

I
Ivan Kokshaysky 已提交
95
EXPORT_SYMBOL(__register_binfmt);
L
Linus Torvalds 已提交
96

97
void unregister_binfmt(struct linux_binfmt * fmt)
L
Linus Torvalds 已提交
98 99
{
	write_lock(&binfmt_lock);
A
Alexey Dobriyan 已提交
100
	list_del(&fmt->lh);
L
Linus Torvalds 已提交
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
	write_unlock(&binfmt_lock);
}

EXPORT_SYMBOL(unregister_binfmt);

static inline void put_binfmt(struct linux_binfmt * fmt)
{
	module_put(fmt->module);
}

/*
 * Note that a shared library must be both readable and executable due to
 * security reasons.
 *
 * Also note that we take the address to load from from the file itself.
 */
117
SYSCALL_DEFINE1(uselib, const char __user *, library)
L
Linus Torvalds 已提交
118
{
119 120 121
	struct file *file;
	char *tmp = getname(library);
	int error = PTR_ERR(tmp);
122 123 124 125 126
	static const struct open_flags uselib_flags = {
		.open_flag = O_LARGEFILE | O_RDONLY | __FMODE_EXEC,
		.acc_mode = MAY_READ | MAY_EXEC | MAY_OPEN,
		.intent = LOOKUP_OPEN
	};
127

128 129 130
	if (IS_ERR(tmp))
		goto out;

131
	file = do_filp_open(AT_FDCWD, tmp, &uselib_flags, LOOKUP_FOLLOW);
132 133 134
	putname(tmp);
	error = PTR_ERR(file);
	if (IS_ERR(file))
L
Linus Torvalds 已提交
135 136 137
		goto out;

	error = -EINVAL;
138
	if (!S_ISREG(file->f_path.dentry->d_inode->i_mode))
L
Linus Torvalds 已提交
139 140
		goto exit;

141
	error = -EACCES;
142
	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
L
Linus Torvalds 已提交
143 144
		goto exit;

145
	fsnotify_open(file);
146

L
Linus Torvalds 已提交
147 148 149 150 151
	error = -ENOEXEC;
	if(file->f_op) {
		struct linux_binfmt * fmt;

		read_lock(&binfmt_lock);
A
Alexey Dobriyan 已提交
152
		list_for_each_entry(fmt, &formats, lh) {
L
Linus Torvalds 已提交
153 154 155 156 157 158 159 160 161 162 163 164 165
			if (!fmt->load_shlib)
				continue;
			if (!try_module_get(fmt->module))
				continue;
			read_unlock(&binfmt_lock);
			error = fmt->load_shlib(file);
			read_lock(&binfmt_lock);
			put_binfmt(fmt);
			if (error != -ENOEXEC)
				break;
		}
		read_unlock(&binfmt_lock);
	}
166
exit:
L
Linus Torvalds 已提交
167 168 169 170 171
	fput(file);
out:
  	return error;
}

172
#ifdef CONFIG_MMU
O
Oleg Nesterov 已提交
173 174 175 176 177 178
/*
 * The nascent bprm->mm is not visible until exec_mmap() but it can
 * use a lot of memory, account these pages in current->mm temporary
 * for oom_badness()->get_mm_rss(). Once exec succeeds or fails, we
 * change the counter back via acct_arg_size(0).
 */
179
static void acct_arg_size(struct linux_binprm *bprm, unsigned long pages)
180 181 182 183 184 185 186 187 188 189 190
{
	struct mm_struct *mm = current->mm;
	long diff = (long)(pages - bprm->vma_pages);

	if (!mm || !diff)
		return;

	bprm->vma_pages = pages;
	add_mm_counter(mm, MM_ANONPAGES, diff);
}

191
static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
192 193 194 195 196 197 198
		int write)
{
	struct page *page;
	int ret;

#ifdef CONFIG_STACK_GROWSUP
	if (write) {
199
		ret = expand_downwards(bprm->vma, pos);
200 201 202 203 204 205 206 207 208 209 210
		if (ret < 0)
			return NULL;
	}
#endif
	ret = get_user_pages(current, bprm->mm, pos,
			1, write, 1, &page, NULL);
	if (ret <= 0)
		return NULL;

	if (write) {
		unsigned long size = bprm->vma->vm_end - bprm->vma->vm_start;
211 212
		struct rlimit *rlim;

213 214
		acct_arg_size(bprm, size / PAGE_SIZE);

215 216 217 218 219 220
		/*
		 * We've historically supported up to 32 pages (ARG_MAX)
		 * of argument strings even with small stacks
		 */
		if (size <= ARG_MAX)
			return page;
221 222 223 224 225 226 227 228

		/*
		 * Limit to 1/4-th the stack size for the argv+env strings.
		 * This ensures that:
		 *  - the remaining binfmt code will not run out of stack space,
		 *  - the program will have a reasonable amount of stack left
		 *    to work from.
		 */
229
		rlim = current->signal->rlim;
J
Jiri Slaby 已提交
230
		if (size > ACCESS_ONCE(rlim[RLIMIT_STACK].rlim_cur) / 4) {
231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
			put_page(page);
			return NULL;
		}
	}

	return page;
}

static void put_arg_page(struct page *page)
{
	put_page(page);
}

static void free_arg_page(struct linux_binprm *bprm, int i)
{
}

static void free_arg_pages(struct linux_binprm *bprm)
{
}

static void flush_arg_page(struct linux_binprm *bprm, unsigned long pos,
		struct page *page)
{
	flush_cache_page(bprm->vma, pos, page_to_pfn(page));
}

static int __bprm_mm_init(struct linux_binprm *bprm)
{
260
	int err;
261 262 263 264 265
	struct vm_area_struct *vma = NULL;
	struct mm_struct *mm = bprm->mm;

	bprm->vma = vma = kmem_cache_zalloc(vm_area_cachep, GFP_KERNEL);
	if (!vma)
266
		return -ENOMEM;
267 268 269 270 271 272 273 274 275 276

	down_write(&mm->mmap_sem);
	vma->vm_mm = mm;

	/*
	 * Place the stack at the largest stack address the architecture
	 * supports. Later, we'll move this to an appropriate place. We don't
	 * use STACK_TOP because that can depend on attributes which aren't
	 * configured yet.
	 */
277
	BUILD_BUG_ON(VM_STACK_FLAGS & VM_STACK_INCOMPLETE_SETUP);
278 279
	vma->vm_end = STACK_TOP_MAX;
	vma->vm_start = vma->vm_end - PAGE_SIZE;
280
	vma->vm_flags = VM_STACK_FLAGS | VM_STACK_INCOMPLETE_SETUP;
281
	vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
282
	INIT_LIST_HEAD(&vma->anon_vma_chain);
283 284 285 286 287

	err = security_file_mmap(NULL, 0, 0, 0, vma->vm_start, 1);
	if (err)
		goto err;

288
	err = insert_vm_struct(mm, vma);
289
	if (err)
290 291 292 293 294 295 296
		goto err;

	mm->stack_vm = mm->total_vm = 1;
	up_write(&mm->mmap_sem);
	bprm->p = vma->vm_end - sizeof(void *);
	return 0;
err:
297 298 299
	up_write(&mm->mmap_sem);
	bprm->vma = NULL;
	kmem_cache_free(vm_area_cachep, vma);
300 301 302 303 304 305 306 307 308 309
	return err;
}

static bool valid_arg_len(struct linux_binprm *bprm, long len)
{
	return len <= MAX_ARG_STRLEN;
}

#else

310
static inline void acct_arg_size(struct linux_binprm *bprm, unsigned long pages)
311 312 313
{
}

314
static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
		int write)
{
	struct page *page;

	page = bprm->page[pos / PAGE_SIZE];
	if (!page && write) {
		page = alloc_page(GFP_HIGHUSER|__GFP_ZERO);
		if (!page)
			return NULL;
		bprm->page[pos / PAGE_SIZE] = page;
	}

	return page;
}

static void put_arg_page(struct page *page)
{
}

static void free_arg_page(struct linux_binprm *bprm, int i)
{
	if (bprm->page[i]) {
		__free_page(bprm->page[i]);
		bprm->page[i] = NULL;
	}
}

static void free_arg_pages(struct linux_binprm *bprm)
{
	int i;

	for (i = 0; i < MAX_ARG_PAGES; i++)
		free_arg_page(bprm, i);
}

static void flush_arg_page(struct linux_binprm *bprm, unsigned long pos,
		struct page *page)
{
}

static int __bprm_mm_init(struct linux_binprm *bprm)
{
	bprm->p = PAGE_SIZE * MAX_ARG_PAGES - sizeof(void *);
	return 0;
}

static bool valid_arg_len(struct linux_binprm *bprm, long len)
{
	return len <= bprm->p;
}

#endif /* CONFIG_MMU */

/*
 * Create a new mm_struct and populate it with a temporary stack
 * vm_area_struct.  We don't have enough context at this point to set the stack
 * flags, permissions, and offset, so we use temporary values.  We'll update
 * them later in setup_arg_pages().
 */
int bprm_mm_init(struct linux_binprm *bprm)
{
	int err;
	struct mm_struct *mm = NULL;

	bprm->mm = mm = mm_alloc();
	err = -ENOMEM;
	if (!mm)
		goto err;

	err = init_new_context(current, mm);
	if (err)
		goto err;

	err = __bprm_mm_init(bprm);
	if (err)
		goto err;

	return 0;

err:
	if (mm) {
		bprm->mm = NULL;
		mmdrop(mm);
	}

	return err;
}

403
struct user_arg_ptr {
404 405 406 407 408 409 410 411 412
#ifdef CONFIG_COMPAT
	bool is_compat;
#endif
	union {
		const char __user *const __user *native;
#ifdef CONFIG_COMPAT
		compat_uptr_t __user *compat;
#endif
	} ptr;
413 414 415
};

static const char __user *get_user_arg_ptr(struct user_arg_ptr argv, int nr)
416
{
417 418 419 420 421 422 423 424
	const char __user *native;

#ifdef CONFIG_COMPAT
	if (unlikely(argv.is_compat)) {
		compat_uptr_t compat;

		if (get_user(compat, argv.ptr.compat + nr))
			return ERR_PTR(-EFAULT);
425

426 427 428 429 430
		return compat_ptr(compat);
	}
#endif

	if (get_user(native, argv.ptr.native + nr))
431 432
		return ERR_PTR(-EFAULT);

433
	return native;
434 435
}

L
Linus Torvalds 已提交
436 437 438
/*
 * count() counts the number of strings in array ARGV.
 */
439
static int count(struct user_arg_ptr argv, int max)
L
Linus Torvalds 已提交
440 441 442
{
	int i = 0;

443
	if (argv.ptr.native != NULL) {
L
Linus Torvalds 已提交
444
		for (;;) {
445
			const char __user *p = get_user_arg_ptr(argv, i);
L
Linus Torvalds 已提交
446 447 448

			if (!p)
				break;
449 450 451 452

			if (IS_ERR(p))
				return -EFAULT;

453
			if (i++ >= max)
L
Linus Torvalds 已提交
454
				return -E2BIG;
455 456 457

			if (fatal_signal_pending(current))
				return -ERESTARTNOHAND;
L
Linus Torvalds 已提交
458 459 460 461 462 463 464
			cond_resched();
		}
	}
	return i;
}

/*
465 466 467
 * 'copy_strings()' copies argument/environment strings from the old
 * processes's memory to the new process's stack.  The call to get_user_pages()
 * ensures the destination page is created and not swapped out.
L
Linus Torvalds 已提交
468
 */
469
static int copy_strings(int argc, struct user_arg_ptr argv,
A
Adrian Bunk 已提交
470
			struct linux_binprm *bprm)
L
Linus Torvalds 已提交
471 472 473
{
	struct page *kmapped_page = NULL;
	char *kaddr = NULL;
474
	unsigned long kpos = 0;
L
Linus Torvalds 已提交
475 476 477
	int ret;

	while (argc-- > 0) {
478
		const char __user *str;
L
Linus Torvalds 已提交
479 480 481
		int len;
		unsigned long pos;

482 483 484
		ret = -EFAULT;
		str = get_user_arg_ptr(argv, argc);
		if (IS_ERR(str))
L
Linus Torvalds 已提交
485 486
			goto out;

487 488 489 490 491 492
		len = strnlen_user(str, MAX_ARG_STRLEN);
		if (!len)
			goto out;

		ret = -E2BIG;
		if (!valid_arg_len(bprm, len))
L
Linus Torvalds 已提交
493 494
			goto out;

495
		/* We're going to work our way backwords. */
L
Linus Torvalds 已提交
496
		pos = bprm->p;
497 498
		str += len;
		bprm->p -= len;
L
Linus Torvalds 已提交
499 500 501 502

		while (len > 0) {
			int offset, bytes_to_copy;

503 504 505 506
			if (fatal_signal_pending(current)) {
				ret = -ERESTARTNOHAND;
				goto out;
			}
507 508
			cond_resched();

L
Linus Torvalds 已提交
509
			offset = pos % PAGE_SIZE;
510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525
			if (offset == 0)
				offset = PAGE_SIZE;

			bytes_to_copy = offset;
			if (bytes_to_copy > len)
				bytes_to_copy = len;

			offset -= bytes_to_copy;
			pos -= bytes_to_copy;
			str -= bytes_to_copy;
			len -= bytes_to_copy;

			if (!kmapped_page || kpos != (pos & PAGE_MASK)) {
				struct page *page;

				page = get_arg_page(bprm, pos, 1);
L
Linus Torvalds 已提交
526
				if (!page) {
527
					ret = -E2BIG;
L
Linus Torvalds 已提交
528 529 530
					goto out;
				}

531 532
				if (kmapped_page) {
					flush_kernel_dcache_page(kmapped_page);
L
Linus Torvalds 已提交
533
					kunmap(kmapped_page);
534 535
					put_arg_page(kmapped_page);
				}
L
Linus Torvalds 已提交
536 537
				kmapped_page = page;
				kaddr = kmap(kmapped_page);
538 539
				kpos = pos & PAGE_MASK;
				flush_arg_page(bprm, kpos, kmapped_page);
L
Linus Torvalds 已提交
540
			}
541
			if (copy_from_user(kaddr+offset, str, bytes_to_copy)) {
L
Linus Torvalds 已提交
542 543 544 545 546 547 548
				ret = -EFAULT;
				goto out;
			}
		}
	}
	ret = 0;
out:
549 550
	if (kmapped_page) {
		flush_kernel_dcache_page(kmapped_page);
L
Linus Torvalds 已提交
551
		kunmap(kmapped_page);
552 553
		put_arg_page(kmapped_page);
	}
L
Linus Torvalds 已提交
554 555 556 557 558 559
	return ret;
}

/*
 * Like copy_strings, but get argv and its values from kernel memory.
 */
560
int copy_strings_kernel(int argc, const char *const *__argv,
561
			struct linux_binprm *bprm)
L
Linus Torvalds 已提交
562 563 564
{
	int r;
	mm_segment_t oldfs = get_fs();
565
	struct user_arg_ptr argv = {
566
		.ptr.native = (const char __user *const  __user *)__argv,
567 568
	};

L
Linus Torvalds 已提交
569
	set_fs(KERNEL_DS);
570
	r = copy_strings(argc, argv, bprm);
L
Linus Torvalds 已提交
571
	set_fs(oldfs);
572

L
Linus Torvalds 已提交
573 574 575 576 577
	return r;
}
EXPORT_SYMBOL(copy_strings_kernel);

#ifdef CONFIG_MMU
578

L
Linus Torvalds 已提交
579
/*
580 581 582
 * During bprm_mm_init(), we create a temporary stack at STACK_TOP_MAX.  Once
 * the binfmt code determines where the new stack should reside, we shift it to
 * its final location.  The process proceeds as follows:
L
Linus Torvalds 已提交
583
 *
584 585 586 587 588 589
 * 1) Use shift to calculate the new vma endpoints.
 * 2) Extend vma to cover both the old and new ranges.  This ensures the
 *    arguments passed to subsequent functions are consistent.
 * 3) Move vma's page tables to the new range.
 * 4) Free up any cleared pgd range.
 * 5) Shrink the vma to cover only the new range.
L
Linus Torvalds 已提交
590
 */
591
static int shift_arg_pages(struct vm_area_struct *vma, unsigned long shift)
L
Linus Torvalds 已提交
592 593
{
	struct mm_struct *mm = vma->vm_mm;
594 595 596 597 598
	unsigned long old_start = vma->vm_start;
	unsigned long old_end = vma->vm_end;
	unsigned long length = old_end - old_start;
	unsigned long new_start = old_start - shift;
	unsigned long new_end = old_end - shift;
P
Peter Zijlstra 已提交
599
	struct mmu_gather tlb;
L
Linus Torvalds 已提交
600

601
	BUG_ON(new_start > new_end);
L
Linus Torvalds 已提交
602

603 604 605 606 607 608 609 610 611 612
	/*
	 * ensure there are no vmas between where we want to go
	 * and where we are
	 */
	if (vma != find_vma(mm, new_start))
		return -EFAULT;

	/*
	 * cover the whole range: [new_start, old_end)
	 */
613 614
	if (vma_adjust(vma, new_start, old_end, vma->vm_pgoff, NULL))
		return -ENOMEM;
615 616 617 618 619 620 621 622 623 624

	/*
	 * move the page tables downwards, on failure we rely on
	 * process cleanup to remove whatever mess we made.
	 */
	if (length != move_page_tables(vma, old_start,
				       vma, new_start, length))
		return -ENOMEM;

	lru_add_drain();
P
Peter Zijlstra 已提交
625
	tlb_gather_mmu(&tlb, mm, 0);
626 627 628 629
	if (new_end > old_start) {
		/*
		 * when the old and new regions overlap clear from new_end.
		 */
P
Peter Zijlstra 已提交
630
		free_pgd_range(&tlb, new_end, old_end, new_end,
631 632 633 634 635 636 637 638
			vma->vm_next ? vma->vm_next->vm_start : 0);
	} else {
		/*
		 * otherwise, clean from old_start; this is done to not touch
		 * the address space in [new_end, old_start) some architectures
		 * have constraints on va-space that make this illegal (IA64) -
		 * for the others its just a little faster.
		 */
P
Peter Zijlstra 已提交
639
		free_pgd_range(&tlb, old_start, old_end, new_end,
640
			vma->vm_next ? vma->vm_next->vm_start : 0);
L
Linus Torvalds 已提交
641
	}
P
Peter Zijlstra 已提交
642
	tlb_finish_mmu(&tlb, new_end, old_end);
643 644

	/*
645
	 * Shrink the vma to just the new range.  Always succeeds.
646 647 648 649
	 */
	vma_adjust(vma, new_start, new_end, vma->vm_pgoff, NULL);

	return 0;
L
Linus Torvalds 已提交
650 651
}

652 653 654 655
/*
 * Finalizes the stack vm_area_struct. The flags and permissions are updated,
 * the stack is optionally relocated, and some extra space is added.
 */
L
Linus Torvalds 已提交
656 657 658 659
int setup_arg_pages(struct linux_binprm *bprm,
		    unsigned long stack_top,
		    int executable_stack)
{
660 661
	unsigned long ret;
	unsigned long stack_shift;
L
Linus Torvalds 已提交
662
	struct mm_struct *mm = current->mm;
663 664 665 666
	struct vm_area_struct *vma = bprm->vma;
	struct vm_area_struct *prev = NULL;
	unsigned long vm_flags;
	unsigned long stack_base;
667 668 669
	unsigned long stack_size;
	unsigned long stack_expand;
	unsigned long rlim_stack;
L
Linus Torvalds 已提交
670 671 672

#ifdef CONFIG_STACK_GROWSUP
	/* Limit stack size to 1GB */
J
Jiri Slaby 已提交
673
	stack_base = rlimit_max(RLIMIT_STACK);
L
Linus Torvalds 已提交
674 675 676
	if (stack_base > (1 << 30))
		stack_base = 1 << 30;

677 678 679
	/* Make sure we didn't let the argument array grow too large. */
	if (vma->vm_end - vma->vm_start > stack_base)
		return -ENOMEM;
L
Linus Torvalds 已提交
680

681
	stack_base = PAGE_ALIGN(stack_top - stack_base);
L
Linus Torvalds 已提交
682

683 684 685
	stack_shift = vma->vm_start - stack_base;
	mm->arg_start = bprm->p - stack_shift;
	bprm->p = vma->vm_end - stack_shift;
L
Linus Torvalds 已提交
686
#else
687 688
	stack_top = arch_align_stack(stack_top);
	stack_top = PAGE_ALIGN(stack_top);
689 690 691 692 693

	if (unlikely(stack_top < mmap_min_addr) ||
	    unlikely(vma->vm_end - vma->vm_start >= stack_top - mmap_min_addr))
		return -ENOMEM;

694 695 696
	stack_shift = vma->vm_end - stack_top;

	bprm->p -= stack_shift;
L
Linus Torvalds 已提交
697 698 699 700
	mm->arg_start = bprm->p;
#endif

	if (bprm->loader)
701 702
		bprm->loader -= stack_shift;
	bprm->exec -= stack_shift;
L
Linus Torvalds 已提交
703 704

	down_write(&mm->mmap_sem);
705
	vm_flags = VM_STACK_FLAGS;
706 707 708 709 710 711 712 713 714 715 716

	/*
	 * Adjust stack execute permissions; explicitly enable for
	 * EXSTACK_ENABLE_X, disable for EXSTACK_DISABLE_X and leave alone
	 * (arch default) otherwise.
	 */
	if (unlikely(executable_stack == EXSTACK_ENABLE_X))
		vm_flags |= VM_EXEC;
	else if (executable_stack == EXSTACK_DISABLE_X)
		vm_flags &= ~VM_EXEC;
	vm_flags |= mm->def_flags;
717
	vm_flags |= VM_STACK_INCOMPLETE_SETUP;
718 719 720 721 722 723 724 725 726 727

	ret = mprotect_fixup(vma, &prev, vma->vm_start, vma->vm_end,
			vm_flags);
	if (ret)
		goto out_unlock;
	BUG_ON(prev != vma);

	/* Move stack pages down in memory. */
	if (stack_shift) {
		ret = shift_arg_pages(vma, stack_shift);
728 729
		if (ret)
			goto out_unlock;
L
Linus Torvalds 已提交
730 731
	}

732 733 734
	/* mprotect_fixup is overkill to remove the temporary stack flags */
	vma->vm_flags &= ~VM_STACK_INCOMPLETE_SETUP;

735
	stack_expand = 131072UL; /* randomly 32*4k (or 2*64k) pages */
736 737 738 739 740 741
	stack_size = vma->vm_end - vma->vm_start;
	/*
	 * Align this down to a page boundary as expand_stack
	 * will align it up.
	 */
	rlim_stack = rlimit(RLIMIT_STACK) & PAGE_MASK;
742
#ifdef CONFIG_STACK_GROWSUP
743 744 745 746
	if (stack_size + stack_expand > rlim_stack)
		stack_base = vma->vm_start + rlim_stack;
	else
		stack_base = vma->vm_end + stack_expand;
747
#else
748 749 750 751
	if (stack_size + stack_expand > rlim_stack)
		stack_base = vma->vm_end - rlim_stack;
	else
		stack_base = vma->vm_start - stack_expand;
752
#endif
753
	current->mm->start_stack = bprm->p;
754 755 756 757 758
	ret = expand_stack(vma, stack_base);
	if (ret)
		ret = -EFAULT;

out_unlock:
L
Linus Torvalds 已提交
759
	up_write(&mm->mmap_sem);
760
	return ret;
L
Linus Torvalds 已提交
761 762 763 764 765 766 767 768
}
EXPORT_SYMBOL(setup_arg_pages);

#endif /* CONFIG_MMU */

struct file *open_exec(const char *name)
{
	struct file *file;
769
	int err;
770 771 772 773 774
	static const struct open_flags open_exec_flags = {
		.open_flag = O_LARGEFILE | O_RDONLY | __FMODE_EXEC,
		.acc_mode = MAY_EXEC | MAY_OPEN,
		.intent = LOOKUP_OPEN
	};
L
Linus Torvalds 已提交
775

776
	file = do_filp_open(AT_FDCWD, name, &open_exec_flags, LOOKUP_FOLLOW);
777
	if (IS_ERR(file))
778 779 780
		goto out;

	err = -EACCES;
781 782
	if (!S_ISREG(file->f_path.dentry->d_inode->i_mode))
		goto exit;
783

784 785
	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
		goto exit;
786

787
	fsnotify_open(file);
788

789
	err = deny_write_access(file);
790 791
	if (err)
		goto exit;
L
Linus Torvalds 已提交
792

793
out:
794 795
	return file;

796 797
exit:
	fput(file);
798 799
	return ERR_PTR(err);
}
L
Linus Torvalds 已提交
800 801
EXPORT_SYMBOL(open_exec);

M
Mimi Zohar 已提交
802 803
int kernel_read(struct file *file, loff_t offset,
		char *addr, unsigned long count)
L
Linus Torvalds 已提交
804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826
{
	mm_segment_t old_fs;
	loff_t pos = offset;
	int result;

	old_fs = get_fs();
	set_fs(get_ds());
	/* The cast to a user pointer is valid due to the set_fs() */
	result = vfs_read(file, (void __user *)addr, count, &pos);
	set_fs(old_fs);
	return result;
}

EXPORT_SYMBOL(kernel_read);

static int exec_mmap(struct mm_struct *mm)
{
	struct task_struct *tsk;
	struct mm_struct * old_mm, *active_mm;

	/* Notify parent that we're no longer interested in the old VM */
	tsk = current;
	old_mm = current->mm;
827
	sync_mm_rss(old_mm);
L
Linus Torvalds 已提交
828 829 830 831 832 833 834
	mm_release(tsk, old_mm);

	if (old_mm) {
		/*
		 * Make sure that if there is a core dump in progress
		 * for the old mm, we get out and die instead of going
		 * through with the exec.  We must hold mmap_sem around
835
		 * checking core_state and changing tsk->mm.
L
Linus Torvalds 已提交
836 837
		 */
		down_read(&old_mm->mmap_sem);
838
		if (unlikely(old_mm->core_state)) {
L
Linus Torvalds 已提交
839 840 841 842 843 844 845 846 847 848 849 850 851
			up_read(&old_mm->mmap_sem);
			return -EINTR;
		}
	}
	task_lock(tsk);
	active_mm = tsk->active_mm;
	tsk->mm = mm;
	tsk->active_mm = mm;
	activate_mm(active_mm, mm);
	task_unlock(tsk);
	arch_pick_mmap_layout(mm);
	if (old_mm) {
		up_read(&old_mm->mmap_sem);
852
		BUG_ON(active_mm != old_mm);
853
		setmax_mm_hiwater_rss(&tsk->signal->maxrss, old_mm);
854
		mm_update_next_owner(old_mm);
L
Linus Torvalds 已提交
855 856 857 858 859 860 861 862 863 864 865 866 867
		mmput(old_mm);
		return 0;
	}
	mmdrop(active_mm);
	return 0;
}

/*
 * This function makes sure the current process has its own signal table,
 * so that flush_signal_handlers can later reset the handlers without
 * disturbing other processes.  (Other processes might share the signal
 * table via the CLONE_SIGHAND option to clone().)
 */
868
static int de_thread(struct task_struct *tsk)
L
Linus Torvalds 已提交
869 870
{
	struct signal_struct *sig = tsk->signal;
871
	struct sighand_struct *oldsighand = tsk->sighand;
L
Linus Torvalds 已提交
872 873
	spinlock_t *lock = &oldsighand->siglock;

874
	if (thread_group_empty(tsk))
L
Linus Torvalds 已提交
875 876 877 878 879 880
		goto no_thread_group;

	/*
	 * Kill all other threads in the thread group.
	 */
	spin_lock_irq(lock);
881
	if (signal_group_exit(sig)) {
L
Linus Torvalds 已提交
882 883 884 885 886 887 888
		/*
		 * Another group action in progress, just
		 * return so that the signal is processed.
		 */
		spin_unlock_irq(lock);
		return -EAGAIN;
	}
889

890
	sig->group_exit_task = tsk;
891 892 893
	sig->notify_count = zap_other_threads(tsk);
	if (!thread_group_leader(tsk))
		sig->notify_count--;
L
Linus Torvalds 已提交
894

895
	while (sig->notify_count) {
L
Linus Torvalds 已提交
896 897 898 899 900 901 902 903 904 905 906 907
		__set_current_state(TASK_UNINTERRUPTIBLE);
		spin_unlock_irq(lock);
		schedule();
		spin_lock_irq(lock);
	}
	spin_unlock_irq(lock);

	/*
	 * At this point all other threads have exited, all we have to
	 * do is to wait for the thread group leader to become inactive,
	 * and to assume its PID:
	 */
908
	if (!thread_group_leader(tsk)) {
909
		struct task_struct *leader = tsk->group_leader;
910

911
		sig->notify_count = -1;	/* for exit_notify() */
912 913 914 915 916 917 918 919
		for (;;) {
			write_lock_irq(&tasklist_lock);
			if (likely(leader->exit_state))
				break;
			__set_current_state(TASK_UNINTERRUPTIBLE);
			write_unlock_irq(&tasklist_lock);
			schedule();
		}
L
Linus Torvalds 已提交
920

921 922 923 924 925 926 927 928 929 930
		/*
		 * The only record we have of the real-time age of a
		 * process, regardless of execs it's done, is start_time.
		 * All the past CPU time is accumulated in signal_struct
		 * from sister threads now dead.  But in this non-leader
		 * exec, nothing survives from the original leader thread,
		 * whose birth marks the true age of this process now.
		 * When we take on its identity by switching to its PID, we
		 * also take its birthdate (always earlier than our own).
		 */
931
		tsk->start_time = leader->start_time;
932

933 934
		BUG_ON(!same_thread_group(leader, tsk));
		BUG_ON(has_group_leader_pid(tsk));
L
Linus Torvalds 已提交
935 936 937 938 939 940
		/*
		 * An exec() starts a new thread group with the
		 * TGID of the previous thread group. Rehash the
		 * two threads with a switched PID, and release
		 * the former thread group leader:
		 */
941 942

		/* Become a process group leader with the old leader's pid.
943 944
		 * The old leader becomes a thread of the this thread group.
		 * Note: The old leader also uses this pid until release_task
945 946
		 *       is called.  Odd but simple and correct.
		 */
947 948
		detach_pid(tsk, PIDTYPE_PID);
		tsk->pid = leader->pid;
949
		attach_pid(tsk, PIDTYPE_PID,  task_pid(leader));
950 951
		transfer_pid(leader, tsk, PIDTYPE_PGID);
		transfer_pid(leader, tsk, PIDTYPE_SID);
952

953
		list_replace_rcu(&leader->tasks, &tsk->tasks);
954
		list_replace_init(&leader->sibling, &tsk->sibling);
L
Linus Torvalds 已提交
955

956 957
		tsk->group_leader = tsk;
		leader->group_leader = tsk;
958

959
		tsk->exit_signal = SIGCHLD;
960
		leader->exit_signal = -1;
961 962 963

		BUG_ON(leader->exit_state != EXIT_ZOMBIE);
		leader->exit_state = EXIT_DEAD;
964 965 966 967 968 969 970 971

		/*
		 * We are going to release_task()->ptrace_unlink() silently,
		 * the tracer can sleep in do_wait(). EXIT_DEAD guarantees
		 * the tracer wont't block again waiting for this thread.
		 */
		if (unlikely(leader->ptrace))
			__wake_up_parent(leader, leader->parent);
L
Linus Torvalds 已提交
972
		write_unlock_irq(&tasklist_lock);
973 974

		release_task(leader);
975
	}
L
Linus Torvalds 已提交
976

977 978
	sig->group_exit_task = NULL;
	sig->notify_count = 0;
L
Linus Torvalds 已提交
979 980

no_thread_group:
981 982 983
	/* we have changed execution domain */
	tsk->exit_signal = SIGCHLD;

L
Linus Torvalds 已提交
984
	exit_itimers(sig);
985
	flush_itimer_signals();
986

987 988
	if (atomic_read(&oldsighand->count) != 1) {
		struct sighand_struct *newsighand;
L
Linus Torvalds 已提交
989
		/*
990 991
		 * This ->sighand is shared with the CLONE_SIGHAND
		 * but not CLONE_THREAD task, switch to the new one.
L
Linus Torvalds 已提交
992
		 */
993 994 995 996
		newsighand = kmem_cache_alloc(sighand_cachep, GFP_KERNEL);
		if (!newsighand)
			return -ENOMEM;

L
Linus Torvalds 已提交
997 998 999 1000 1001 1002
		atomic_set(&newsighand->count, 1);
		memcpy(newsighand->action, oldsighand->action,
		       sizeof(newsighand->action));

		write_lock_irq(&tasklist_lock);
		spin_lock(&oldsighand->siglock);
1003
		rcu_assign_pointer(tsk->sighand, newsighand);
L
Linus Torvalds 已提交
1004 1005 1006
		spin_unlock(&oldsighand->siglock);
		write_unlock_irq(&tasklist_lock);

1007
		__cleanup_sighand(oldsighand);
L
Linus Torvalds 已提交
1008 1009
	}

1010
	BUG_ON(!thread_group_leader(tsk));
L
Linus Torvalds 已提交
1011 1012
	return 0;
}
O
Oleg Nesterov 已提交
1013

L
Linus Torvalds 已提交
1014 1015 1016 1017
/*
 * These functions flushes out all traces of the currently running executable
 * so that a new one can be started
 */
1018
static void flush_old_files(struct files_struct * files)
L
Linus Torvalds 已提交
1019 1020
{
	long j = -1;
1021
	struct fdtable *fdt;
L
Linus Torvalds 已提交
1022 1023 1024 1025 1026 1027 1028

	spin_lock(&files->file_lock);
	for (;;) {
		unsigned long set, i;

		j++;
		i = j * __NFDBITS;
1029
		fdt = files_fdtable(files);
1030
		if (i >= fdt->max_fds)
L
Linus Torvalds 已提交
1031
			break;
1032
		set = fdt->close_on_exec->fds_bits[j];
L
Linus Torvalds 已提交
1033 1034
		if (!set)
			continue;
1035
		fdt->close_on_exec->fds_bits[j] = 0;
L
Linus Torvalds 已提交
1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
		spin_unlock(&files->file_lock);
		for ( ; set ; i++,set >>= 1) {
			if (set & 1) {
				sys_close(i);
			}
		}
		spin_lock(&files->file_lock);

	}
	spin_unlock(&files->file_lock);
}

1048
char *get_task_comm(char *buf, struct task_struct *tsk)
L
Linus Torvalds 已提交
1049 1050 1051 1052 1053
{
	/* buf must be at least sizeof(tsk->comm) in size */
	task_lock(tsk);
	strncpy(buf, tsk->comm, sizeof(tsk->comm));
	task_unlock(tsk);
1054
	return buf;
L
Linus Torvalds 已提交
1055
}
1056
EXPORT_SYMBOL_GPL(get_task_comm);
L
Linus Torvalds 已提交
1057 1058 1059 1060

void set_task_comm(struct task_struct *tsk, char *buf)
{
	task_lock(tsk);
1061

1062 1063
	trace_task_rename(tsk, buf);

1064 1065 1066 1067 1068 1069 1070 1071
	/*
	 * Threads may access current->comm without holding
	 * the task lock, so write the string carefully.
	 * Readers without a lock may see incomplete new
	 * names but are safe from non-terminating string reads.
	 */
	memset(tsk->comm, 0, TASK_COMM_LEN);
	wmb();
L
Linus Torvalds 已提交
1072 1073
	strlcpy(tsk->comm, buf, sizeof(tsk->comm));
	task_unlock(tsk);
1074
	perf_event_comm(tsk);
L
Linus Torvalds 已提交
1075 1076
}

1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
static void filename_to_taskname(char *tcomm, const char *fn, unsigned int len)
{
	int i, ch;

	/* Copies the binary name from after last slash */
	for (i = 0; (ch = *(fn++)) != '\0';) {
		if (ch == '/')
			i = 0; /* overwrite what we wrote */
		else
			if (i < len - 1)
				tcomm[i++] = ch;
	}
	tcomm[i] = '\0';
}

L
Linus Torvalds 已提交
1092 1093
int flush_old_exec(struct linux_binprm * bprm)
{
1094
	int retval;
L
Linus Torvalds 已提交
1095 1096 1097 1098 1099 1100 1101 1102 1103

	/*
	 * Make sure we have a private signal table and that
	 * we are unassociated from the previous thread group.
	 */
	retval = de_thread(current);
	if (retval)
		goto out;

M
Matt Helsley 已提交
1104 1105
	set_mm_exe_file(bprm->mm, bprm->file);

1106
	filename_to_taskname(bprm->tcomm, bprm->filename, sizeof(bprm->tcomm));
L
Linus Torvalds 已提交
1107 1108 1109
	/*
	 * Release all of the old mmap stuff
	 */
1110
	acct_arg_size(bprm, 0);
L
Linus Torvalds 已提交
1111 1112
	retval = exec_mmap(bprm->mm);
	if (retval)
1113
		goto out;
L
Linus Torvalds 已提交
1114 1115

	bprm->mm = NULL;		/* We're using it now */
1116

1117
	set_fs(USER_DS);
1118
	current->flags &= ~(PF_RANDOMIZE | PF_KTHREAD);
1119 1120 1121
	flush_thread();
	current->personality &= ~bprm->per_clear;

1122 1123 1124 1125 1126 1127 1128
	return 0;

out:
	return retval;
}
EXPORT_SYMBOL(flush_old_exec);

1129 1130 1131 1132 1133 1134 1135
void would_dump(struct linux_binprm *bprm, struct file *file)
{
	if (inode_permission(file->f_path.dentry->d_inode, MAY_READ) < 0)
		bprm->interp_flags |= BINPRM_FLAGS_ENFORCE_NONDUMP;
}
EXPORT_SYMBOL(would_dump);

1136 1137 1138
void setup_new_exec(struct linux_binprm * bprm)
{
	arch_pick_mmap_layout(current->mm);
L
Linus Torvalds 已提交
1139 1140 1141 1142

	/* This is the point of no return */
	current->sas_ss_sp = current->sas_ss_size = 0;

1143
	if (current_euid() == current_uid() && current_egid() == current_gid())
1144
		set_dumpable(current->mm, 1);
A
Alan Cox 已提交
1145
	else
1146
		set_dumpable(current->mm, suid_dumpable);
A
Alan Cox 已提交
1147

1148
	set_task_comm(current, bprm->tcomm);
L
Linus Torvalds 已提交
1149

1150 1151 1152 1153 1154 1155
	/* Set the new mm task size. We have to do that late because it may
	 * depend on TIF_32BIT which is only updated in flush_thread() on
	 * some architectures like powerpc
	 */
	current->mm->task_size = TASK_SIZE;

1156 1157 1158
	/* install the new credentials */
	if (bprm->cred->uid != current_euid() ||
	    bprm->cred->gid != current_egid()) {
1159
		current->pdeath_signal = 0;
1160 1161 1162 1163
	} else {
		would_dump(bprm, bprm->file);
		if (bprm->interp_flags & BINPRM_FLAGS_ENFORCE_NONDUMP)
			set_dumpable(current->mm, suid_dumpable);
L
Linus Torvalds 已提交
1164 1165
	}

I
Ingo Molnar 已提交
1166 1167 1168 1169 1170
	/*
	 * Flush performance counters when crossing a
	 * security domain:
	 */
	if (!get_dumpable(current->mm))
1171
		perf_event_exit_task(current);
I
Ingo Molnar 已提交
1172

L
Linus Torvalds 已提交
1173 1174 1175 1176 1177 1178 1179 1180
	/* An exec changes our domain. We are no longer part of the thread
	   group */

	current->self_exec_id++;
			
	flush_signal_handlers(current, 0);
	flush_old_files(current->files);
}
1181
EXPORT_SYMBOL(setup_new_exec);
L
Linus Torvalds 已提交
1182

1183 1184 1185 1186 1187 1188 1189 1190
/*
 * Prepare credentials and lock ->cred_guard_mutex.
 * install_exec_creds() commits the new creds and drops the lock.
 * Or, if exec fails before, free_bprm() should release ->cred and
 * and unlock.
 */
int prepare_bprm_creds(struct linux_binprm *bprm)
{
1191
	if (mutex_lock_interruptible(&current->signal->cred_guard_mutex))
1192 1193 1194 1195 1196 1197
		return -ERESTARTNOINTR;

	bprm->cred = prepare_exec_creds();
	if (likely(bprm->cred))
		return 0;

1198
	mutex_unlock(&current->signal->cred_guard_mutex);
1199 1200 1201 1202 1203 1204 1205
	return -ENOMEM;
}

void free_bprm(struct linux_binprm *bprm)
{
	free_arg_pages(bprm);
	if (bprm->cred) {
1206
		mutex_unlock(&current->signal->cred_guard_mutex);
1207 1208 1209 1210 1211
		abort_creds(bprm->cred);
	}
	kfree(bprm);
}

1212 1213 1214 1215 1216 1217 1218 1219 1220
/*
 * install the new credentials for this executable
 */
void install_exec_creds(struct linux_binprm *bprm)
{
	security_bprm_committing_creds(bprm);

	commit_creds(bprm->cred);
	bprm->cred = NULL;
1221 1222
	/*
	 * cred_guard_mutex must be held at least to this point to prevent
1223
	 * ptrace_attach() from altering our determination of the task's
1224 1225
	 * credentials; any time after this it may be unlocked.
	 */
1226
	security_bprm_committed_creds(bprm);
1227
	mutex_unlock(&current->signal->cred_guard_mutex);
1228 1229 1230 1231 1232
}
EXPORT_SYMBOL(install_exec_creds);

/*
 * determine how safe it is to execute the proposed program
1233
 * - the caller must hold ->cred_guard_mutex to protect against
1234 1235
 *   PTRACE_ATTACH
 */
A
Al Viro 已提交
1236
static int check_unsafe_exec(struct linux_binprm *bprm)
1237
{
D
David Howells 已提交
1238
	struct task_struct *p = current, *t;
1239
	unsigned n_fs;
A
Al Viro 已提交
1240
	int res = 0;
1241

T
Tejun Heo 已提交
1242 1243 1244 1245 1246 1247
	if (p->ptrace) {
		if (p->ptrace & PT_PTRACE_CAP)
			bprm->unsafe |= LSM_UNSAFE_PTRACE_CAP;
		else
			bprm->unsafe |= LSM_UNSAFE_PTRACE;
	}
1248

D
David Howells 已提交
1249
	n_fs = 1;
N
Nick Piggin 已提交
1250
	spin_lock(&p->fs->lock);
1251
	rcu_read_lock();
D
David Howells 已提交
1252 1253 1254 1255
	for (t = next_thread(p); t != p; t = next_thread(t)) {
		if (t->fs == p->fs)
			n_fs++;
	}
1256
	rcu_read_unlock();
D
David Howells 已提交
1257

1258
	if (p->fs->users > n_fs) {
1259
		bprm->unsafe |= LSM_UNSAFE_SHARE;
A
Al Viro 已提交
1260
	} else {
1261 1262 1263 1264 1265
		res = -EAGAIN;
		if (!p->fs->in_exec) {
			p->fs->in_exec = 1;
			res = 1;
		}
A
Al Viro 已提交
1266
	}
N
Nick Piggin 已提交
1267
	spin_unlock(&p->fs->lock);
A
Al Viro 已提交
1268 1269

	return res;
1270 1271
}

L
Linus Torvalds 已提交
1272 1273 1274
/* 
 * Fill the binprm structure from the inode. 
 * Check permissions, then read the first 128 (BINPRM_BUF_SIZE) bytes
1275 1276
 *
 * This may be called multiple times for binary chains (scripts for example).
L
Linus Torvalds 已提交
1277 1278 1279
 */
int prepare_binprm(struct linux_binprm *bprm)
{
1280
	umode_t mode;
1281
	struct inode * inode = bprm->file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1282 1283 1284 1285 1286 1287
	int retval;

	mode = inode->i_mode;
	if (bprm->file->f_op == NULL)
		return -EACCES;

1288 1289 1290
	/* clear any previous set[ug]id data from a previous binary */
	bprm->cred->euid = current_euid();
	bprm->cred->egid = current_egid();
L
Linus Torvalds 已提交
1291

1292
	if (!(bprm->file->f_path.mnt->mnt_flags & MNT_NOSUID)) {
L
Linus Torvalds 已提交
1293 1294
		/* Set-uid? */
		if (mode & S_ISUID) {
1295 1296
			bprm->per_clear |= PER_CLEAR_ON_SETID;
			bprm->cred->euid = inode->i_uid;
L
Linus Torvalds 已提交
1297 1298 1299 1300 1301 1302 1303 1304 1305
		}

		/* Set-gid? */
		/*
		 * If setgid is set but no group execute bit then this
		 * is a candidate for mandatory locking, not a setgid
		 * executable.
		 */
		if ((mode & (S_ISGID | S_IXGRP)) == (S_ISGID | S_IXGRP)) {
1306 1307
			bprm->per_clear |= PER_CLEAR_ON_SETID;
			bprm->cred->egid = inode->i_gid;
L
Linus Torvalds 已提交
1308 1309 1310 1311
		}
	}

	/* fill in binprm security blob */
1312
	retval = security_bprm_set_creds(bprm);
L
Linus Torvalds 已提交
1313 1314
	if (retval)
		return retval;
1315
	bprm->cred_prepared = 1;
L
Linus Torvalds 已提交
1316

1317 1318
	memset(bprm->buf, 0, BINPRM_BUF_SIZE);
	return kernel_read(bprm->file, 0, bprm->buf, BINPRM_BUF_SIZE);
L
Linus Torvalds 已提交
1319 1320 1321 1322
}

EXPORT_SYMBOL(prepare_binprm);

N
Nick Piggin 已提交
1323 1324 1325 1326 1327
/*
 * Arguments are '\0' separated strings found at the location bprm->p
 * points to; chop off the first by relocating brpm->p to right after
 * the first '\0' encountered.
 */
1328
int remove_arg_zero(struct linux_binprm *bprm)
L
Linus Torvalds 已提交
1329
{
1330 1331 1332 1333
	int ret = 0;
	unsigned long offset;
	char *kaddr;
	struct page *page;
N
Nick Piggin 已提交
1334

1335 1336
	if (!bprm->argc)
		return 0;
L
Linus Torvalds 已提交
1337

1338 1339 1340 1341 1342 1343 1344
	do {
		offset = bprm->p & ~PAGE_MASK;
		page = get_arg_page(bprm, bprm->p, 0);
		if (!page) {
			ret = -EFAULT;
			goto out;
		}
1345
		kaddr = kmap_atomic(page);
N
Nick Piggin 已提交
1346

1347 1348 1349
		for (; offset < PAGE_SIZE && kaddr[offset];
				offset++, bprm->p++)
			;
N
Nick Piggin 已提交
1350

1351
		kunmap_atomic(kaddr);
1352
		put_arg_page(page);
N
Nick Piggin 已提交
1353

1354 1355 1356
		if (offset == PAGE_SIZE)
			free_arg_page(bprm, (bprm->p >> PAGE_SHIFT) - 1);
	} while (offset == PAGE_SIZE);
N
Nick Piggin 已提交
1357

1358 1359 1360
	bprm->p++;
	bprm->argc--;
	ret = 0;
N
Nick Piggin 已提交
1361

1362 1363
out:
	return ret;
L
Linus Torvalds 已提交
1364 1365 1366 1367 1368 1369 1370 1371
}
EXPORT_SYMBOL(remove_arg_zero);

/*
 * cycle the list of binary formats handler, until one recognizes the image
 */
int search_binary_handler(struct linux_binprm *bprm,struct pt_regs *regs)
{
1372
	unsigned int depth = bprm->recursion_depth;
L
Linus Torvalds 已提交
1373 1374
	int try,retval;
	struct linux_binfmt *fmt;
1375
	pid_t old_pid;
L
Linus Torvalds 已提交
1376 1377 1378 1379 1380

	retval = security_bprm_check(bprm);
	if (retval)
		return retval;

A
Al Viro 已提交
1381 1382 1383 1384
	retval = audit_bprm(bprm);
	if (retval)
		return retval;

1385 1386 1387 1388 1389
	/* Need to fetch pid before load_binary changes it */
	rcu_read_lock();
	old_pid = task_pid_nr_ns(current, task_active_pid_ns(current->parent));
	rcu_read_unlock();

L
Linus Torvalds 已提交
1390 1391 1392
	retval = -ENOENT;
	for (try=0; try<2; try++) {
		read_lock(&binfmt_lock);
A
Alexey Dobriyan 已提交
1393
		list_for_each_entry(fmt, &formats, lh) {
L
Linus Torvalds 已提交
1394 1395 1396 1397 1398 1399 1400
			int (*fn)(struct linux_binprm *, struct pt_regs *) = fmt->load_binary;
			if (!fn)
				continue;
			if (!try_module_get(fmt->module))
				continue;
			read_unlock(&binfmt_lock);
			retval = fn(bprm, regs);
1401 1402 1403 1404 1405 1406
			/*
			 * Restore the depth counter to its starting value
			 * in this call, so we don't have to rely on every
			 * load_binary function to restore it on return.
			 */
			bprm->recursion_depth = depth;
L
Linus Torvalds 已提交
1407
			if (retval >= 0) {
1408 1409 1410 1411
				if (depth == 0) {
					trace_sched_process_exec(current, old_pid, bprm);
					ptrace_event(PTRACE_EVENT_EXEC, old_pid);
				}
L
Linus Torvalds 已提交
1412 1413 1414 1415 1416 1417
				put_binfmt(fmt);
				allow_write_access(bprm->file);
				if (bprm->file)
					fput(bprm->file);
				bprm->file = NULL;
				current->did_exec = 1;
M
Matt Helsley 已提交
1418
				proc_exec_connector(current);
L
Linus Torvalds 已提交
1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430
				return retval;
			}
			read_lock(&binfmt_lock);
			put_binfmt(fmt);
			if (retval != -ENOEXEC || bprm->mm == NULL)
				break;
			if (!bprm->file) {
				read_unlock(&binfmt_lock);
				return retval;
			}
		}
		read_unlock(&binfmt_lock);
1431
#ifdef CONFIG_MODULES
L
Linus Torvalds 已提交
1432 1433
		if (retval != -ENOEXEC || bprm->mm == NULL) {
			break;
J
Johannes Berg 已提交
1434
		} else {
L
Linus Torvalds 已提交
1435 1436 1437 1438 1439 1440
#define printable(c) (((c)=='\t') || ((c)=='\n') || (0x20<=(c) && (c)<=0x7e))
			if (printable(bprm->buf[0]) &&
			    printable(bprm->buf[1]) &&
			    printable(bprm->buf[2]) &&
			    printable(bprm->buf[3]))
				break; /* -ENOEXEC */
1441 1442
			if (try)
				break; /* -ENOEXEC */
L
Linus Torvalds 已提交
1443 1444
			request_module("binfmt-%04x", *(unsigned short *)(&bprm->buf[2]));
		}
1445 1446 1447
#else
		break;
#endif
L
Linus Torvalds 已提交
1448 1449 1450 1451 1452 1453 1454 1455 1456
	}
	return retval;
}

EXPORT_SYMBOL(search_binary_handler);

/*
 * sys_execve() executes a new program.
 */
1457 1458 1459 1460
static int do_execve_common(const char *filename,
				struct user_arg_ptr argv,
				struct user_arg_ptr envp,
				struct pt_regs *regs)
L
Linus Torvalds 已提交
1461 1462 1463
{
	struct linux_binprm *bprm;
	struct file *file;
1464
	struct files_struct *displaced;
1465
	bool clear_in_exec;
L
Linus Torvalds 已提交
1466
	int retval;
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
	const struct cred *cred = current_cred();

	/*
	 * We move the actual failure in case of RLIMIT_NPROC excess from
	 * set*uid() to execve() because too many poorly written programs
	 * don't check setuid() return code.  Here we additionally recheck
	 * whether NPROC limit is still exceeded.
	 */
	if ((current->flags & PF_NPROC_EXCEEDED) &&
	    atomic_read(&cred->user->processes) > rlimit(RLIMIT_NPROC)) {
		retval = -EAGAIN;
		goto out_ret;
	}

	/* We're below the limit (still or again), so we don't want to make
	 * further execve() calls fail. */
	current->flags &= ~PF_NPROC_EXCEEDED;
L
Linus Torvalds 已提交
1484

1485
	retval = unshare_files(&displaced);
1486 1487 1488
	if (retval)
		goto out_ret;

L
Linus Torvalds 已提交
1489
	retval = -ENOMEM;
1490
	bprm = kzalloc(sizeof(*bprm), GFP_KERNEL);
L
Linus Torvalds 已提交
1491
	if (!bprm)
1492
		goto out_files;
L
Linus Torvalds 已提交
1493

1494 1495
	retval = prepare_bprm_creds(bprm);
	if (retval)
1496
		goto out_free;
A
Al Viro 已提交
1497 1498

	retval = check_unsafe_exec(bprm);
1499
	if (retval < 0)
1500
		goto out_free;
1501
	clear_in_exec = retval;
1502
	current->in_execve = 1;
1503

L
Linus Torvalds 已提交
1504 1505 1506
	file = open_exec(filename);
	retval = PTR_ERR(file);
	if (IS_ERR(file))
A
Al Viro 已提交
1507
		goto out_unmark;
L
Linus Torvalds 已提交
1508 1509 1510 1511 1512 1513 1514

	sched_exec();

	bprm->file = file;
	bprm->filename = filename;
	bprm->interp = filename;

1515 1516 1517
	retval = bprm_mm_init(bprm);
	if (retval)
		goto out_file;
L
Linus Torvalds 已提交
1518

1519
	bprm->argc = count(argv, MAX_ARG_STRINGS);
L
Linus Torvalds 已提交
1520
	if ((retval = bprm->argc) < 0)
1521
		goto out;
L
Linus Torvalds 已提交
1522

1523
	bprm->envc = count(envp, MAX_ARG_STRINGS);
L
Linus Torvalds 已提交
1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544
	if ((retval = bprm->envc) < 0)
		goto out;

	retval = prepare_binprm(bprm);
	if (retval < 0)
		goto out;

	retval = copy_strings_kernel(1, &bprm->filename, bprm);
	if (retval < 0)
		goto out;

	bprm->exec = bprm->p;
	retval = copy_strings(bprm->envc, envp, bprm);
	if (retval < 0)
		goto out;

	retval = copy_strings(bprm->argc, argv, bprm);
	if (retval < 0)
		goto out;

	retval = search_binary_handler(bprm,regs);
1545 1546
	if (retval < 0)
		goto out;
L
Linus Torvalds 已提交
1547

1548
	/* execve succeeded */
A
Al Viro 已提交
1549
	current->fs->in_exec = 0;
1550
	current->in_execve = 0;
1551 1552 1553 1554 1555
	acct_update_integrals(current);
	free_bprm(bprm);
	if (displaced)
		put_files_struct(displaced);
	return retval;
L
Linus Torvalds 已提交
1556

1557
out:
1558 1559 1560 1561
	if (bprm->mm) {
		acct_arg_size(bprm, 0);
		mmput(bprm->mm);
	}
L
Linus Torvalds 已提交
1562 1563 1564 1565 1566 1567

out_file:
	if (bprm->file) {
		allow_write_access(bprm->file);
		fput(bprm->file);
	}
1568

A
Al Viro 已提交
1569
out_unmark:
1570 1571
	if (clear_in_exec)
		current->fs->in_exec = 0;
1572
	current->in_execve = 0;
1573 1574

out_free:
1575
	free_bprm(bprm);
L
Linus Torvalds 已提交
1576

1577
out_files:
1578 1579
	if (displaced)
		reset_files_struct(displaced);
L
Linus Torvalds 已提交
1580 1581 1582 1583
out_ret:
	return retval;
}

1584 1585 1586 1587 1588
int do_execve(const char *filename,
	const char __user *const __user *__argv,
	const char __user *const __user *__envp,
	struct pt_regs *regs)
{
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
	struct user_arg_ptr argv = { .ptr.native = __argv };
	struct user_arg_ptr envp = { .ptr.native = __envp };
	return do_execve_common(filename, argv, envp, regs);
}

#ifdef CONFIG_COMPAT
int compat_do_execve(char *filename,
	compat_uptr_t __user *__argv,
	compat_uptr_t __user *__envp,
	struct pt_regs *regs)
{
	struct user_arg_ptr argv = {
		.is_compat = true,
		.ptr.compat = __argv,
	};
	struct user_arg_ptr envp = {
		.is_compat = true,
		.ptr.compat = __envp,
	};
1608 1609
	return do_execve_common(filename, argv, envp, regs);
}
1610
#endif
1611

1612
void set_binfmt(struct linux_binfmt *new)
L
Linus Torvalds 已提交
1613
{
1614 1615 1616 1617
	struct mm_struct *mm = current->mm;

	if (mm->binfmt)
		module_put(mm->binfmt->module);
L
Linus Torvalds 已提交
1618

1619
	mm->binfmt = new;
1620 1621
	if (new)
		__module_get(new->module);
L
Linus Torvalds 已提交
1622 1623 1624 1625
}

EXPORT_SYMBOL(set_binfmt);

1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
static int expand_corename(struct core_name *cn)
{
	char *old_corename = cn->corename;

	cn->size = CORENAME_MAX_SIZE * atomic_inc_return(&call_count);
	cn->corename = krealloc(old_corename, cn->size, GFP_KERNEL);

	if (!cn->corename) {
		kfree(old_corename);
		return -ENOMEM;
	}

	return 0;
}

static int cn_printf(struct core_name *cn, const char *fmt, ...)
{
	char *cur;
	int need;
	int ret;
	va_list arg;

	va_start(arg, fmt);
	need = vsnprintf(NULL, 0, fmt, arg);
	va_end(arg);

	if (likely(need < cn->size - cn->used - 1))
		goto out_printf;

	ret = expand_corename(cn);
	if (ret)
		goto expand_fail;

out_printf:
	cur = cn->corename + cn->used;
	va_start(arg, fmt);
	vsnprintf(cur, need + 1, fmt, arg);
	va_end(arg);
	cn->used += need;
	return 0;

expand_fail:
	return ret;
}

1671 1672 1673 1674 1675 1676 1677
static void cn_escape(char *str)
{
	for (; *str; str++)
		if (*str == '/')
			*str = '!';
}

1678 1679 1680
static int cn_print_exe_file(struct core_name *cn)
{
	struct file *exe_file;
1681
	char *pathbuf, *path;
1682 1683 1684
	int ret;

	exe_file = get_mm_exe_file(current->mm);
1685 1686 1687 1688 1689 1690
	if (!exe_file) {
		char *commstart = cn->corename + cn->used;
		ret = cn_printf(cn, "%s (path unknown)", current->comm);
		cn_escape(commstart);
		return ret;
	}
1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703

	pathbuf = kmalloc(PATH_MAX, GFP_TEMPORARY);
	if (!pathbuf) {
		ret = -ENOMEM;
		goto put_exe_file;
	}

	path = d_path(&exe_file->f_path, pathbuf, PATH_MAX);
	if (IS_ERR(path)) {
		ret = PTR_ERR(path);
		goto free_buf;
	}

1704
	cn_escape(path);
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714

	ret = cn_printf(cn, "%s", path);

free_buf:
	kfree(pathbuf);
put_exe_file:
	fput(exe_file);
	return ret;
}

L
Linus Torvalds 已提交
1715 1716 1717 1718
/* format_corename will inspect the pattern parameter, and output a
 * name into corename, which must have space for at least
 * CORENAME_MAX_SIZE bytes plus one byte for the zero terminator.
 */
1719
static int format_corename(struct core_name *cn, long signr)
L
Linus Torvalds 已提交
1720
{
1721
	const struct cred *cred = current_cred();
1722 1723
	const char *pat_ptr = core_pattern;
	int ispipe = (*pat_ptr == '|');
L
Linus Torvalds 已提交
1724
	int pid_in_pattern = 0;
1725 1726 1727 1728 1729 1730 1731 1732
	int err = 0;

	cn->size = CORENAME_MAX_SIZE * atomic_read(&call_count);
	cn->corename = kmalloc(cn->size, GFP_KERNEL);
	cn->used = 0;

	if (!cn->corename)
		return -ENOMEM;
L
Linus Torvalds 已提交
1733 1734 1735 1736 1737

	/* Repeat as long as we have more pattern to process and more output
	   space */
	while (*pat_ptr) {
		if (*pat_ptr != '%') {
1738
			if (*pat_ptr == 0)
L
Linus Torvalds 已提交
1739
				goto out;
1740
			err = cn_printf(cn, "%c", *pat_ptr++);
L
Linus Torvalds 已提交
1741 1742
		} else {
			switch (*++pat_ptr) {
1743
			/* single % at the end, drop that */
L
Linus Torvalds 已提交
1744 1745 1746 1747
			case 0:
				goto out;
			/* Double percent, output one percent */
			case '%':
1748
				err = cn_printf(cn, "%c", '%');
L
Linus Torvalds 已提交
1749 1750 1751 1752
				break;
			/* pid */
			case 'p':
				pid_in_pattern = 1;
1753 1754
				err = cn_printf(cn, "%d",
					      task_tgid_vnr(current));
L
Linus Torvalds 已提交
1755 1756 1757
				break;
			/* uid */
			case 'u':
1758
				err = cn_printf(cn, "%d", cred->uid);
L
Linus Torvalds 已提交
1759 1760 1761
				break;
			/* gid */
			case 'g':
1762
				err = cn_printf(cn, "%d", cred->gid);
L
Linus Torvalds 已提交
1763 1764 1765
				break;
			/* signal that caused the coredump */
			case 's':
1766
				err = cn_printf(cn, "%ld", signr);
L
Linus Torvalds 已提交
1767 1768 1769 1770 1771
				break;
			/* UNIX time of coredump */
			case 't': {
				struct timeval tv;
				do_gettimeofday(&tv);
1772
				err = cn_printf(cn, "%lu", tv.tv_sec);
L
Linus Torvalds 已提交
1773 1774 1775
				break;
			}
			/* hostname */
1776 1777
			case 'h': {
				char *namestart = cn->corename + cn->used;
L
Linus Torvalds 已提交
1778
				down_read(&uts_sem);
1779 1780
				err = cn_printf(cn, "%s",
					      utsname()->nodename);
L
Linus Torvalds 已提交
1781
				up_read(&uts_sem);
1782
				cn_escape(namestart);
L
Linus Torvalds 已提交
1783
				break;
1784
			}
L
Linus Torvalds 已提交
1785
			/* executable */
1786 1787
			case 'e': {
				char *commstart = cn->corename + cn->used;
1788
				err = cn_printf(cn, "%s", current->comm);
1789
				cn_escape(commstart);
L
Linus Torvalds 已提交
1790
				break;
1791
			}
1792 1793 1794
			case 'E':
				err = cn_print_exe_file(cn);
				break;
1795 1796
			/* core limit size */
			case 'c':
1797 1798
				err = cn_printf(cn, "%lu",
					      rlimit(RLIMIT_CORE));
1799
				break;
L
Linus Torvalds 已提交
1800 1801 1802 1803 1804
			default:
				break;
			}
			++pat_ptr;
		}
1805 1806 1807

		if (err)
			return err;
L
Linus Torvalds 已提交
1808
	}
1809

L
Linus Torvalds 已提交
1810 1811 1812 1813
	/* Backward compatibility with core_uses_pid:
	 *
	 * If core_pattern does not include a %p (as is the default)
	 * and core_uses_pid is set, then .%pid will be appended to
1814
	 * the filename. Do not do this for piped commands. */
1815
	if (!ispipe && !pid_in_pattern && core_uses_pid) {
1816 1817 1818
		err = cn_printf(cn, ".%d", task_tgid_vnr(current));
		if (err)
			return err;
L
Linus Torvalds 已提交
1819
	}
1820 1821
out:
	return ispipe;
L
Linus Torvalds 已提交
1822 1823
}

1824
static int zap_process(struct task_struct *start, int exit_code)
1825 1826
{
	struct task_struct *t;
1827
	int nr = 0;
1828

1829
	start->signal->flags = SIGNAL_GROUP_EXIT;
1830
	start->signal->group_exit_code = exit_code;
1831
	start->signal->group_stop_count = 0;
1832 1833 1834

	t = start;
	do {
1835
		task_clear_jobctl_pending(t, JOBCTL_PENDING_MASK);
1836
		if (t != current && t->mm) {
1837 1838
			sigaddset(&t->pending.signal, SIGKILL);
			signal_wake_up(t, 1);
1839
			nr++;
1840
		}
1841
	} while_each_thread(start, t);
1842 1843

	return nr;
1844 1845
}

1846
static inline int zap_threads(struct task_struct *tsk, struct mm_struct *mm,
1847
				struct core_state *core_state, int exit_code)
L
Linus Torvalds 已提交
1848 1849
{
	struct task_struct *g, *p;
1850
	unsigned long flags;
1851
	int nr = -EAGAIN;
1852 1853

	spin_lock_irq(&tsk->sighand->siglock);
1854
	if (!signal_group_exit(tsk->signal)) {
1855
		mm->core_state = core_state;
1856
		nr = zap_process(tsk, exit_code);
L
Linus Torvalds 已提交
1857
	}
1858
	spin_unlock_irq(&tsk->sighand->siglock);
1859 1860
	if (unlikely(nr < 0))
		return nr;
L
Linus Torvalds 已提交
1861

1862
	if (atomic_read(&mm->mm_users) == nr + 1)
1863
		goto done;
1864 1865
	/*
	 * We should find and kill all tasks which use this mm, and we should
1866
	 * count them correctly into ->nr_threads. We don't take tasklist
1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893
	 * lock, but this is safe wrt:
	 *
	 * fork:
	 *	None of sub-threads can fork after zap_process(leader). All
	 *	processes which were created before this point should be
	 *	visible to zap_threads() because copy_process() adds the new
	 *	process to the tail of init_task.tasks list, and lock/unlock
	 *	of ->siglock provides a memory barrier.
	 *
	 * do_exit:
	 *	The caller holds mm->mmap_sem. This means that the task which
	 *	uses this mm can't pass exit_mm(), so it can't exit or clear
	 *	its ->mm.
	 *
	 * de_thread:
	 *	It does list_replace_rcu(&leader->tasks, &current->tasks),
	 *	we must see either old or new leader, this does not matter.
	 *	However, it can change p->sighand, so lock_task_sighand(p)
	 *	must be used. Since p->mm != NULL and we hold ->mmap_sem
	 *	it can't fail.
	 *
	 *	Note also that "g" can be the old leader with ->mm == NULL
	 *	and already unhashed and thus removed from ->thread_group.
	 *	This is OK, __unhash_process()->list_del_rcu() does not
	 *	clear the ->next pointer, we will find the new leader via
	 *	next_thread().
	 */
1894
	rcu_read_lock();
1895
	for_each_process(g) {
1896 1897
		if (g == tsk->group_leader)
			continue;
1898 1899
		if (g->flags & PF_KTHREAD)
			continue;
1900 1901 1902
		p = g;
		do {
			if (p->mm) {
1903
				if (unlikely(p->mm == mm)) {
1904
					lock_task_sighand(p, &flags);
1905
					nr += zap_process(p, exit_code);
1906 1907
					unlock_task_sighand(p, &flags);
				}
1908 1909
				break;
			}
1910
		} while_each_thread(g, p);
1911
	}
1912
	rcu_read_unlock();
1913
done:
1914
	atomic_set(&core_state->nr_threads, nr);
1915
	return nr;
L
Linus Torvalds 已提交
1916 1917
}

1918
static int coredump_wait(int exit_code, struct core_state *core_state)
L
Linus Torvalds 已提交
1919
{
1920 1921
	struct task_struct *tsk = current;
	struct mm_struct *mm = tsk->mm;
1922
	int core_waiters = -EBUSY;
L
Linus Torvalds 已提交
1923

1924
	init_completion(&core_state->startup);
1925 1926
	core_state->dumper.task = tsk;
	core_state->dumper.next = NULL;
1927 1928 1929 1930

	down_write(&mm->mmap_sem);
	if (!mm->core_state)
		core_waiters = zap_threads(tsk, mm, core_state, exit_code);
O
Oleg Nesterov 已提交
1931 1932
	up_write(&mm->mmap_sem);

1933
	if (core_waiters > 0)
1934
		wait_for_completion(&core_state->startup);
1935

1936
	return core_waiters;
L
Linus Torvalds 已提交
1937 1938
}

O
Oleg Nesterov 已提交
1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
static void coredump_finish(struct mm_struct *mm)
{
	struct core_thread *curr, *next;
	struct task_struct *task;

	next = mm->core_state->dumper.next;
	while ((curr = next) != NULL) {
		next = curr->next;
		task = curr->task;
		/*
		 * see exit_mm(), curr->task must not see
		 * ->task == NULL before we read ->next.
		 */
		smp_mb();
		curr->task = NULL;
		wake_up_process(task);
	}

	mm->core_state = NULL;
}

1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
/*
 * set_dumpable converts traditional three-value dumpable to two flags and
 * stores them into mm->flags.  It modifies lower two bits of mm->flags, but
 * these bits are not changed atomically.  So get_dumpable can observe the
 * intermediate state.  To avoid doing unexpected behavior, get get_dumpable
 * return either old dumpable or new one by paying attention to the order of
 * modifying the bits.
 *
 * dumpable |   mm->flags (binary)
 * old  new | initial interim  final
 * ---------+-----------------------
 *  0    1  |   00      01      01
 *  0    2  |   00      10(*)   11
 *  1    0  |   01      00      00
 *  1    2  |   01      11      11
 *  2    0  |   11      10(*)   00
 *  2    1  |   11      11      01
 *
 * (*) get_dumpable regards interim value of 10 as 11.
 */
void set_dumpable(struct mm_struct *mm, int value)
{
	switch (value) {
	case 0:
		clear_bit(MMF_DUMPABLE, &mm->flags);
		smp_wmb();
		clear_bit(MMF_DUMP_SECURELY, &mm->flags);
		break;
	case 1:
		set_bit(MMF_DUMPABLE, &mm->flags);
		smp_wmb();
		clear_bit(MMF_DUMP_SECURELY, &mm->flags);
		break;
	case 2:
		set_bit(MMF_DUMP_SECURELY, &mm->flags);
		smp_wmb();
		set_bit(MMF_DUMPABLE, &mm->flags);
		break;
	}
}

2001
static int __get_dumpable(unsigned long mm_flags)
2002 2003 2004
{
	int ret;

2005
	ret = mm_flags & MMF_DUMPABLE_MASK;
2006 2007 2008
	return (ret >= 2) ? 2 : ret;
}

2009 2010 2011 2012 2013
int get_dumpable(struct mm_struct *mm)
{
	return __get_dumpable(mm->flags);
}

2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036
static void wait_for_dump_helpers(struct file *file)
{
	struct pipe_inode_info *pipe;

	pipe = file->f_path.dentry->d_inode->i_pipe;

	pipe_lock(pipe);
	pipe->readers++;
	pipe->writers--;

	while ((pipe->readers > 1) && (!signal_pending(current))) {
		wake_up_interruptible_sync(&pipe->wait);
		kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
		pipe_wait(pipe);
	}

	pipe->readers--;
	pipe->writers++;
	pipe_unlock(pipe);

}


2037
/*
H
Holger Hans Peter Freyther 已提交
2038
 * umh_pipe_setup
2039 2040 2041 2042 2043 2044 2045 2046 2047
 * helper function to customize the process used
 * to collect the core in userspace.  Specifically
 * it sets up a pipe and installs it as fd 0 (stdin)
 * for the process.  Returns 0 on success, or
 * PTR_ERR on failure.
 * Note that it also sets the core limit to 1.  This
 * is a special value that we use to trap recursive
 * core dumps
 */
2048
static int umh_pipe_setup(struct subprocess_info *info, struct cred *new)
2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080
{
	struct file *rp, *wp;
	struct fdtable *fdt;
	struct coredump_params *cp = (struct coredump_params *)info->data;
	struct files_struct *cf = current->files;

	wp = create_write_pipe(0);
	if (IS_ERR(wp))
		return PTR_ERR(wp);

	rp = create_read_pipe(wp, 0);
	if (IS_ERR(rp)) {
		free_write_pipe(wp);
		return PTR_ERR(rp);
	}

	cp->file = wp;

	sys_close(0);
	fd_install(0, rp);
	spin_lock(&cf->file_lock);
	fdt = files_fdtable(cf);
	FD_SET(0, fdt->open_fds);
	FD_CLR(0, fdt->close_on_exec);
	spin_unlock(&cf->file_lock);

	/* and disallow core files too */
	current->signal->rlim[RLIMIT_CORE] = (struct rlimit){1, 1};

	return 0;
}

W
WANG Cong 已提交
2081
void do_coredump(long signr, int exit_code, struct pt_regs *regs)
L
Linus Torvalds 已提交
2082
{
2083
	struct core_state core_state;
2084
	struct core_name cn;
L
Linus Torvalds 已提交
2085 2086
	struct mm_struct *mm = current->mm;
	struct linux_binfmt * binfmt;
D
David Howells 已提交
2087 2088
	const struct cred *old_cred;
	struct cred *cred;
L
Linus Torvalds 已提交
2089
	int retval = 0;
A
Alan Cox 已提交
2090
	int flag = 0;
O
Oleg Nesterov 已提交
2091
	int ispipe;
2092
	static atomic_t core_dump_count = ATOMIC_INIT(0);
2093 2094 2095
	struct coredump_params cprm = {
		.signr = signr,
		.regs = regs,
J
Jiri Slaby 已提交
2096
		.limit = rlimit(RLIMIT_CORE),
2097 2098 2099 2100 2101 2102
		/*
		 * We must use the same mm->flags while dumping core to avoid
		 * inconsistency of bit flags, since this flag is not protected
		 * by any locks.
		 */
		.mm_flags = mm->flags,
2103
	};
L
Linus Torvalds 已提交
2104

S
Steve Grubb 已提交
2105 2106
	audit_core_dumps(signr);

2107
	binfmt = mm->binfmt;
L
Linus Torvalds 已提交
2108 2109
	if (!binfmt || !binfmt->core_dump)
		goto fail;
2110 2111
	if (!__get_dumpable(cprm.mm_flags))
		goto fail;
D
David Howells 已提交
2112 2113

	cred = prepare_creds();
2114
	if (!cred)
D
David Howells 已提交
2115
		goto fail;
A
Alan Cox 已提交
2116 2117 2118 2119 2120
	/*
	 *	We cannot trust fsuid as being the "true" uid of the
	 *	process nor do we know its entire history. We only know it
	 *	was tainted so we dump it as root in mode 2.
	 */
2121 2122
	if (__get_dumpable(cprm.mm_flags) == 2) {
		/* Setuid core dump mode */
A
Alan Cox 已提交
2123
		flag = O_EXCL;		/* Stop rewrite attacks */
D
David Howells 已提交
2124
		cred->fsuid = 0;	/* Dump root private */
A
Alan Cox 已提交
2125
	}
2126

2127
	retval = coredump_wait(exit_code, &core_state);
2128 2129
	if (retval < 0)
		goto fail_creds;
D
David Howells 已提交
2130 2131

	old_cred = override_creds(cred);
L
Linus Torvalds 已提交
2132 2133 2134 2135 2136 2137 2138

	/*
	 * Clear any false indication of pending signals that might
	 * be seen by the filesystem code called to write the core file.
	 */
	clear_thread_flag(TIF_SIGPENDING);

2139 2140
	ispipe = format_corename(&cn, signr);

2141
 	if (ispipe) {
O
Oleg Nesterov 已提交
2142 2143 2144
		int dump_count;
		char **helper_argv;

2145 2146 2147 2148 2149 2150
		if (ispipe < 0) {
			printk(KERN_WARNING "format_corename failed\n");
			printk(KERN_WARNING "Aborting core\n");
			goto fail_corename;
		}

2151
		if (cprm.limit == 1) {
2152 2153 2154
			/*
			 * Normally core limits are irrelevant to pipes, since
			 * we're not writing to the file system, but we use
2155 2156
			 * cprm.limit of 1 here as a speacial value. Any
			 * non-1 limit gets set to RLIM_INFINITY below, but
2157 2158
			 * a limit of 0 skips the dump.  This is a consistent
			 * way to catch recursive crashes.  We can still crash
2159
			 * if the core_pattern binary sets RLIM_CORE =  !1
2160 2161 2162 2163 2164 2165 2166
			 * but it runs as root, and can do lots of stupid things
			 * Note that we use task_tgid_vnr here to grab the pid
			 * of the process group leader.  That way we get the
			 * right pid if a thread in a multi-threaded
			 * core_pattern process dies.
			 */
			printk(KERN_WARNING
2167
				"Process %d(%s) has RLIMIT_CORE set to 1\n",
2168 2169 2170 2171
				task_tgid_vnr(current), current->comm);
			printk(KERN_WARNING "Aborting core\n");
			goto fail_unlock;
		}
O
Oleg Nesterov 已提交
2172
		cprm.limit = RLIM_INFINITY;
2173

2174 2175 2176 2177 2178 2179 2180 2181
		dump_count = atomic_inc_return(&core_dump_count);
		if (core_pipe_limit && (core_pipe_limit < dump_count)) {
			printk(KERN_WARNING "Pid %d(%s) over core_pipe_limit\n",
			       task_tgid_vnr(current), current->comm);
			printk(KERN_WARNING "Skipping core dump\n");
			goto fail_dropcount;
		}

2182
		helper_argv = argv_split(GFP_KERNEL, cn.corename+1, NULL);
2183 2184 2185
		if (!helper_argv) {
			printk(KERN_WARNING "%s failed to allocate memory\n",
			       __func__);
2186
			goto fail_dropcount;
2187
		}
2188

O
Oleg Nesterov 已提交
2189 2190 2191 2192 2193
		retval = call_usermodehelper_fns(helper_argv[0], helper_argv,
					NULL, UMH_WAIT_EXEC, umh_pipe_setup,
					NULL, &cprm);
		argv_free(helper_argv);
		if (retval) {
2194
 			printk(KERN_INFO "Core dump to %s pipe failed\n",
2195
			       cn.corename);
O
Oleg Nesterov 已提交
2196
			goto close_fail;
2197
 		}
2198 2199 2200 2201 2202 2203
	} else {
		struct inode *inode;

		if (cprm.limit < binfmt->min_coredump)
			goto fail_unlock;

2204
		cprm.file = filp_open(cn.corename,
2205 2206
				 O_CREAT | 2 | O_NOFOLLOW | O_LARGEFILE | flag,
				 0600);
2207 2208
		if (IS_ERR(cprm.file))
			goto fail_unlock;
L
Linus Torvalds 已提交
2209

2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231
		inode = cprm.file->f_path.dentry->d_inode;
		if (inode->i_nlink > 1)
			goto close_fail;
		if (d_unhashed(cprm.file->f_path.dentry))
			goto close_fail;
		/*
		 * AK: actually i see no reason to not allow this for named
		 * pipes etc, but keep the previous behaviour for now.
		 */
		if (!S_ISREG(inode->i_mode))
			goto close_fail;
		/*
		 * Dont allow local users get cute and trick others to coredump
		 * into their pre-created files.
		 */
		if (inode->i_uid != current_fsuid())
			goto close_fail;
		if (!cprm.file->f_op || !cprm.file->f_op->write)
			goto close_fail;
		if (do_truncate(cprm.file->f_path.dentry, 0, 0, cprm.file))
			goto close_fail;
	}
L
Linus Torvalds 已提交
2232

2233
	retval = binfmt->core_dump(&cprm);
L
Linus Torvalds 已提交
2234 2235
	if (retval)
		current->signal->group_exit_code |= 0x80;
O
Oleg Nesterov 已提交
2236

2237
	if (ispipe && core_pipe_limit)
2238
		wait_for_dump_helpers(cprm.file);
O
Oleg Nesterov 已提交
2239 2240 2241
close_fail:
	if (cprm.file)
		filp_close(cprm.file, NULL);
2242
fail_dropcount:
O
Oleg Nesterov 已提交
2243
	if (ispipe)
2244
		atomic_dec(&core_dump_count);
L
Linus Torvalds 已提交
2245
fail_unlock:
2246 2247
	kfree(cn.corename);
fail_corename:
2248
	coredump_finish(mm);
D
David Howells 已提交
2249
	revert_creds(old_cred);
2250
fail_creds:
D
David Howells 已提交
2251
	put_cred(cred);
L
Linus Torvalds 已提交
2252
fail:
W
WANG Cong 已提交
2253
	return;
L
Linus Torvalds 已提交
2254
}
2255 2256 2257 2258 2259 2260 2261 2262 2263 2264

/*
 * Core dumping helper functions.  These are the only things you should
 * do on a core-file: use only these functions to write out all the
 * necessary info.
 */
int dump_write(struct file *file, const void *addr, int nr)
{
	return access_ok(VERIFY_READ, addr, nr) && file->f_op->write(file, addr, nr, &file->f_pos) == nr;
}
2265
EXPORT_SYMBOL(dump_write);
2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293

int dump_seek(struct file *file, loff_t off)
{
	int ret = 1;

	if (file->f_op->llseek && file->f_op->llseek != no_llseek) {
		if (file->f_op->llseek(file, off, SEEK_CUR) < 0)
			return 0;
	} else {
		char *buf = (char *)get_zeroed_page(GFP_KERNEL);

		if (!buf)
			return 0;
		while (off > 0) {
			unsigned long n = off;

			if (n > PAGE_SIZE)
				n = PAGE_SIZE;
			if (!dump_write(file, buf, n)) {
				ret = 0;
				break;
			}
			off -= n;
		}
		free_page((unsigned long)buf);
	}
	return ret;
}
2294
EXPORT_SYMBOL(dump_seek);